Daniel's been thinking about hostage situations — the ones in movies, the ones in true crime series — and wondering whether they're actually as common as pop culture makes them seem. He says he doesn't know anyone who's ever been taken hostage, suspects most home burglaries and bank robberies don't involve hostages at all, and wants to know the numbers. He's also asking how much the picture changes between countries with robust policing and those with weaker systems. The question came out of a research rabbit hole where he hit a bunch of FBI 404 errors, which turns out to be part of the story.
The 404 errors are absolutely part of the story. The FBI used to maintain something called HOBAS — the Hostage Barricade Database System — and it was the gold standard for this kind of data in the United States. Incident reports, outcomes, negotiation tactics, the works. And then around 2020 it just... went dark. The public-facing portal started returning 404s, and nobody's really explained why. So Daniel's frustration there is genuine, and it's actually the first thing you need to know about hostage statistics: the best dataset we had is no longer publicly accessible.
A database about hostage situations being held hostage by the FBI. There's a joke in there somewhere.
But the data vacuum matters because it forces researchers to rely on academic studies and fragmented sources. The UNODC — the UN Office on Drugs and Crime — has its own crime statistics portal, but their International Classification of Crime for Statistical Purposes, the ICCS, doesn't even have a dedicated code for hostage situations. It gets lumped under kidnapping or deprivation of liberty. So from the start, we're dealing with a definitional mess. A domestic dispute where someone holds a family member at knifepoint for twenty minutes and a coordinated terrorist siege lasting three days both get called hostage situations, but they're not the same thing at all.
So before we even get to Daniel's question about how common these are, we have to figure out what we're counting.
And the cinematic template doesn't help. Hollywood gives us one archetype: armed robber takes civilians, negotiator talks, SWAT breaches. It's a stranger-on-stranger event, it's in a public place, and it unfolds over hours with a ticking clock. That image is so baked into the culture that when people hear "hostage situation," that's what they picture. But the reality is messier, shorter, and often doesn't involve strangers at all.
Daniel's instinct that he doesn't know anyone who's been taken hostage — that's probably accurate for the Hollywood version. But the broader category might be closer to home than he thinks.
Right. The academic consensus — and this comes largely from work by people like Dr. Michael Gelles, who was the chief psychologist for the Naval Criminal Investigative Service — is that roughly sixty to seventy percent of what gets classified as a hostage or barricade incident in the US is actually a domestic dispute or a suicide-by-cop scenario. Someone barricades themselves with a family member. It's not a bank robbery gone wrong. It's not a stranger grabbing someone off the street. It's intimate partner violence that escalates to the point where someone is being held against their will and police are called.
So the most common hostage situation in America is a guy who's already in crisis, already known to the people in the house, and the "hostage" is his wife or his kid.
That's the modal case. And it usually resolves in under two hours. The dramatic sixteen-hour standoff with a negotiator on the phone is the outlier, not the norm. Most of these are over before the SWAT team even finishes setting up.
Which makes the Hollywood version not just inaccurate but inverted. The thing we think of as the typical hostage situation is actually the weird edge case.
And Daniel's specific intuitions hold up really well. He mentioned bank robberies — are those ever actually hostage situations? FBI data from the 2010s showed that fewer than four percent of bank robberies involved any hostage-taking. The vast majority are note jobs or verbal demands, grab the cash, out the door in under two minutes. Taking a hostage adds complexity and risk that most bank robbers actively want to avoid. It turns a property crime into a potential federal kidnapping charge. It brings in negotiators, SWAT, the whole apparatus. Bank robbers are not, as a rule, strategic masterminds, but most of them understand that hostages are a liability.
Four percent. So Daniel's right. The bank robbery with hostages is basically a movie trope.
It's a movie trope, and the numbers bear that out. Home invasions are even more lopsided. The FBI's National Incident-Based Reporting System — NIBRS — doesn't even have a dedicated code for "hostage during burglary." That tells you how rare it's considered. When a hostage situation does occur during a residential burglary, it's almost always opportunistic. The homeowner comes home unexpectedly, the burglar panics and improvises. It's not a planned kidnapping. It's a burglary that went sideways.
So far Daniel's batting a thousand. Bank robberies, rare. Home invasions, rare. Not knowing anyone who's been taken hostage, statistically normal.
And yet — the total number of hostage and barricade incidents in the US is not zero. Depending on which academic estimate you use, it's somewhere in the range of a few hundred to a couple thousand per year. The problem is that without HOBAS, we can't pin that number down precisely. What we can say is that the vast majority of those incidents don't look like the movies. They're domestic. They're short. And they don't make national news.
A few hundred to a couple thousand. For context, the US has about three hundred thirty million people. So your odds of being in a hostage situation in any given year are... what, roughly one in a hundred thousand? Something like that?
On the order of one in a hundred thousand, yeah. Maybe even lower depending on which estimate you trust. For comparison, your lifetime odds of being struck by lightning are about one in fifteen thousand. So Daniel's intuition that this is a rare event is completely correct for the kind of hostage situation he's imagining.
He's more likely to be struck by lightning than taken hostage in a bank robbery. That feels like a useful calibration.
But here's where the story gets interesting, and where the second part of Daniel's question comes in. What happens when you look at countries with weaker policing?
Right. Because everything we've just described is the picture in a country where the state is functional, police respond quickly, and the legal system has teeth. What happens when you remove those assumptions?
The gradient is stark. The UNODC data on kidnapping — which, remember, is the closest proxy we have for hostage-taking in international statistics — shows that countries with high police effectiveness ratings report kidnapping rates below one per hundred thousand. Japan, Singapore, Germany. Countries with weak or corrupt policing report rates ten to fifty times higher. Mexico, Brazil, South Africa. We're not talking about a subtle difference. We're talking about orders of magnitude.
Give me some concrete numbers.
Japan's National Police Agency reported twelve kidnapping cases nationwide in 2023. Twelve. In a country of a hundred twenty-five million people. And of those twelve, zero involved barricaded standoffs with police. Zero. Now contrast that with Mexico. In 2022, the Mexican government recorded one thousand seventy-four kidnappings. And the dark figure — the number of cases that go unreported — is estimated by researchers at five to ten times higher. So the real number could be north of five thousand, maybe ten thousand.
Twelve versus potentially ten thousand. That's not a difference of degree. That's a completely different phenomenon.
And it's not just the frequency. The nature of the crime changes fundamentally. In strong-state environments, hostage-taking is almost always expressive. It's emotional, it's domestic, it's a mental health crisis. The person taking the hostage isn't doing it as a rational criminal strategy. They're in crisis, they're desperate, they're not thinking clearly. In weak-state environments, hostage-taking becomes instrumental. It's a tool of organized crime. It's a business.
Say more about that distinction. Expressive versus instrumental.
An expressive hostage event is someone who's lost control and is acting out of emotion. The hostage is a symbol, a way to exert power in a moment of crisis. The outcome isn't about money or leverage — it's about being heard, being feared, being taken seriously. An instrumental hostage event is a transaction. The hostage is an asset. You take them because someone will pay to get them back, or because you're sending a message to a rival, or because you're collecting a debt. The hostage has a price.
And in Mexico or Brazil, that's the dominant form.
It is. Mexican cartels use what they call levantones — pick-ups — as a standard enforcement mechanism. Someone owes money, you pick them up, you hold them until the debt is paid. It's not even considered a separate category of crime internally. It's just how business gets done. In Brazil, the PCC — Primeiro Comando da Capital, the largest criminal organization in the country — runs hostage-taking as a service. You pay them to kidnap someone who owes you money. It's outsourced debt collection.
So hostage-taking becomes a market instrument. Supply and demand.
And that changes everything about how police respond. In a strong-state environment, the default police response to a hostage situation is negotiation. You assume the hostage-taker is a person in crisis, you try to de-escalate, you bring in a negotiator, and tactical resolution is the last resort. In a weak-state environment, negotiation often doesn't make sense because you're not dealing with someone in crisis. You're dealing with a rational actor who may kill the hostage the moment police show up, because the hostage has already served their purpose or because killing them eliminates a witness. The calculus is completely different.
So the policing strategy that works in Connecticut doesn't work in Sinaloa.
It doesn't. And the really uncomfortable implication is that the nature of hostage-taking is a mirror of state capacity. Where the state is strong, hostages are a symptom of individual crisis. Where the state is weak, hostages are a market instrument. The same act — holding someone against their will — means completely different things depending on the institutional environment around it.
That's the thing Daniel was really poking at, I think. He asked about the difference between robust and less robust policing systems, but the answer isn't just "more hostages in weak states." It's that the entire category changes meaning.
And we haven't even talked about the cases where the state is somewhere in the middle. Countries where policing is present but corrupt, or where the state has effective control in cities but not in rural areas. The Philippines, parts of India, parts of sub-Saharan Africa. In those environments you get a hybrid: some hostage events are expressive, some are instrumental, and the police response is inconsistent because the police themselves may be part of the problem.
There's a case from last year that illustrates this. Seventeen migrants kidnapped in northern Mexico by the Gulf Cartel. Held for six days. Resolved by a ransom payment — the families paid. The police were never informed. That's not a failure of policing. That's a situation where calling the police was never even on the table.
Because in that environment, the police might be working with the cartel, or they might be so outgunned that involving them makes things worse. The families made a rational calculation: pay the ransom, get your people back, don't bring in the state. That's what instrumental hostage-taking looks like in practice. It's a private transaction between criminals and victims, and the state is irrelevant.
Contrast that with the 2014 Sydney Lindt Cafe siege. Lone actor, no prior criminal record, seventeen-hour standoff, two hostages killed. That's closer to the Hollywood template, but it's statistically anomalous. A lone wolf with a grievance, not a criminal enterprise. The police negotiated for hours. The whole thing was broadcast live. And the outcome was still tragic, but the state was fully engaged the entire time.
The Sydney siege is actually a useful case because it shows what happens when the Hollywood template meets reality. The public expects a dramatic resolution. The police are under enormous pressure to act. And the reality is that most hostage situations end quietly, or they end badly, and the dramatic SWAT breach that saves everyone is mostly fiction.
The North Hollywood shootout in 1997 is another one that gets misremembered. People think of it as a hostage event, but the robbers never took hostages. They just shot at police for forty-four minutes. The public memory conflates "barricaded gunman" with "hostage taker," but they're different things.
And that conflation is everywhere. The FBI's data categories actually distinguish between barricade situations, where someone is holed up alone, and hostage situations, where someone is holding another person. But the media almost never makes that distinction, and neither do most people. So when someone says "hostage situation," they might be picturing a barricaded gunman, or a domestic dispute, or a terrorist siege, or a cartel kidnapping. The word is doing too much work.
Which brings us back to the data problem. If we can't even agree on what we're counting, how do we compare across countries?
We do it badly. The UNODC's ICCS doesn't have a dedicated hostage situation code. It gets filed under kidnapping or deprivation of liberty. So when Japan reports twelve kidnappings and Mexico reports over a thousand, we're comparing numbers that were generated by completely different classification systems, different reporting standards, different levels of public trust in police. The Japanese number is probably accurate. The Mexican number is almost certainly an undercount, maybe a massive one. And neither number tells you how many of those cases involved barricades, negotiations, or tactical resolutions.
So the cross-country comparison is built on shaky foundations, but the gradient is too stark to ignore. Even if the Mexican number is off by a factor of five, it's still orders of magnitude higher than Japan.
That's the thing. You can quibble with the methodology, but the pattern holds across dozens of countries. Strong state capacity, low kidnapping rates. Weak state capacity, high kidnapping rates. And the mechanism isn't mysterious. If you think you can get away with it, and the payoff is high enough, someone will do it. Hostage-taking is just a particularly extreme version of that logic.
Daniel also mentioned advances in bank security. I think he was wondering whether those have pushed hostage situations down over time.
They probably have, though it's hard to prove causally. Bank robberies in the US have been declining for decades. In 1990 there were over nine thousand. By 2020 it was down to about fifteen hundred. Better cameras, dye packs, tracking devices, less cash on hand — the risk-reward calculus shifted. And since bank robberies were never a major source of hostage situations to begin with, the decline in bank robberies probably didn't move the needle much on hostage statistics. But it did make the movie version even less realistic.
So the Hollywood bank robbery hostage scene is a double fiction. Bank robberies are rare, and hostage-taking during bank robberies is rarer still.
Yet the image persists because it's dramatically useful. A hostage creates stakes. It gives the negotiator something to do. It gives the SWAT team a reason to stack up outside. It's a narrtive engine. The fact that it barely happens in real life is almost beside the point for a screenwriter.
There's something almost comforting about the movie version, isn't there? The idea that when someone takes a hostage, a calm professional shows up and talks them down. The system works. The state is competent. The reality in most of the world is that if you're taken hostage, the state may never know, and if it does know, it may not be able to help.
That's the institutional mirror Daniel was asking about. Hostage statistics don't just tell you about crime. They tell you about whether people trust the state enough to report crimes, whether the police are capable of responding, whether the legal system can punish offenders. A low kidnapping rate might mean the country is safe. Or it might mean nobody bothers to call the police because they know nothing will happen.
Which is the dark figure problem. The unreported cases. In Mexico, researchers estimate the real kidnapping number is five to ten times the official figure. That's not a measurement error. That's a signal about institutional trust.
It cuts both ways. In a country like Japan, the kidnapping rate is low partly because the state is effective and partly because the social norms make certain kinds of crime unthinkable. You can't separate the policing from the culture. In a country like Brazil, the kidnapping rate is high partly because the state is weak and partly because organized crime has filled the vacuum. The PCC runs kidnapping as a service because the state can't stop them and because there's demand for it.
The answer to Daniel's question — "am I underestimating how common these are?" — is that it depends entirely on where you are and what you mean by hostage situation. In the United States, the Hollywood version is rare. The domestic dispute version is more common but still not something most people will ever encounter. In Mexico or Brazil, the instrumental version is frighteningly common, but it doesn't look like the movies either. It looks like a business transaction.
That's the thing I want to sit with. The gap between the cinematic hostage event and the real one isn't just about frequency. It's about meaning. In the movies, a hostage situation is a crisis of individual morality — will the negotiator say the right thing, will the hostage-taker pull the trigger. In real life, it's a reflection of institutional capacity. The question isn't "will the hero save the day." It's "does the state work."
Hilbert: I fell asleep during one.
...What?
Hilbert: Late nineties. I was working for a police training firm in the UK. They'd hire actors to play hostage-takers in simulated scenarios. Mock bank, fake gun, headset. Real negotiators on the other end of the line, trying to talk you down. The pay was forty quid a session.
You were a crisis negotiator roleplayer.
Hilbert: That's a generous title. I sat in a room with a plastic gun and a cold pizza and waited for someone to call. Sometimes it took hours. The real skill — and I watched enough actual negotiators to know this — the real skill isn't the dramatic phone call. It's the patience to sit in silence for forty-five minutes while someone on the other end works up the courage to say something stupid.
And you fell asleep.
Hilbert: Third session of the day. The room was warm. I'd been up late the night before. I nodded off in the chair. The trainee negotiator on the other end thought I'd gone silent as a tactical move. Wrote in his assessment that I was testing his patience. He passed.
He thought you were playing mind games and you were just unconscious.
Hilbert: Best evaluation I ever gave. Didn't say a word for twenty minutes and got marked as advanced.
That's... a perfect summary of the gap between the movies and reality. The negotiator thought he was in a psychological thriller, and you were taking a nap.
Hilbert: Most of the scenarios ended like that. Not the nap part — the boredom part. Someone shouting for the first ten minutes, then hours of nothing. The negotiator trying to build rapport while you're counting ceiling tiles. The dramatic breach almost never happened because the trainers wanted the negotiators to practice patience, not tactics.
The training was designed around the reality, not the movie version.
Hilbert: The training was designed around the fact that most hostage-takers are exhausted, scared, and not thinking clearly. You don't need to outsmart them. You need to outwait them. The phone call is just a way to pass the time until they're too tired to keep holding the gun.
How long did you do this?
Hilbert: About a year. The firm lost its Home Office contract and I went back to driving a van. But I still think about it whenever I see a hostage scene in a film. All that tension, the ticking clock, the hero negotiator saying exactly the right thing at exactly the right moment. Real life is a man in a chair, trying not to fall asleep, while someone on the phone asks him for the fourth time what he wants to eat.
The cold pizza detail is going to stick with me.
Hilbert: It was cheese and tomato. I remember because I'd been staring at it for three hours and the negotiator kept asking if I needed anything and I kept saying no because I wasn't sure if asking for a fresh pizza would break character.
The theatrical integrity of the hostage roleplayer.
Hilbert: You don't break the scenario. That was the one rule. Whatever you do, stay in the room. Even if you're asleep, apparently.
The fact that the trainee passed because he interpreted your silence as a tactic — that's actually a good sign for his training. He was learning to sit with uncertainty, to not fill every silence. That's what good negotiators do.
Hilbert: He was learning the right thing for the wrong reason. Which is probably true of most training, now I think about it.
The question Daniel's prompt leaves me with — and I think this is where we should land — is what happens as policing systems in the Global South become more professionalized over time. If hostage-taking is a mirror of state capacity, do we see a shift from instrumental to expressive hostage events as the state gets stronger? Or does the criminal market adapt faster than the state can build capacity?
There's some evidence for both. In Colombia, kidnapping rates dropped dramatically after the 2016 peace deal with the FARC, but the criminal groups that filled the vacuum adapted quickly. They shifted from political kidnapping to extortion and micro-trafficking. The market evolved. In Mexico, despite some professionalization of federal police, the kidnapping rate hasn't moved much because the cartels have diversified so thoroughly that kidnapping is just one revenue stream among many.
The state gets better, but the criminals get better too.
That's the arms race. And the question is whether the state can get better faster than the criminals can adapt. In some places, the answer seems to be yes — Chile and Uruguay have seen real declines in kidnapping over the past two decades as their institutions have strengthened. In other places, the criminals are so far ahead that catching up will take generations.
Daniel's intuition was mostly right. The hostage situations he imagines are rare. But the category is bigger and stranger than he thought, and the real story is about what a society's hostage statistics say about its institutions. The next time you see a hostage scene in a movie, ask yourself: what kind of state does this society have? Because the answer changes everything about whether that scene makes any sense.
That's the thing I'll take from this. The movies show us a crisis of individual character. The data shows us a crisis of institutional capacity. And the gap between those two things is where most of the interesting questions live.
Thanks to Hilbert Flumingtop for producing, and for spending a year of his life eating cold pizza in a fake bank.
This has been My Weird Prompts. If you want to send us a question — maybe one that doesn't involve hostage statistics, though we're not picky — you can reach us at show at my weird prompts dot com.
We'll be back soon. Until then, don't take any hostages, and if you do, at least order fresh pizza.
